balls in play and how they pertain to the statistics BABIP, FIP, and ERA

[SirKodiak]

More by Eric Seidman on "The Zambrano/Bonderman Conundrum":
http://www.fangraphs.com/blogs/index...rman-conundrum

Here is the mailbag answer by MGL that Seidman refers to in the article:
http://tangotiger.net/wiki/index.php...why_the_gap.3F

[dps]

I certainly don't want to close my mind to learning something new, but it does still read to me like instead of changing his mind about FIP when someone proves FIP's weakness (Zambrano), he's instead trying to imply the theory is fine and Zambrano is just the exception, when in my mind the theory is wrong and therefore the stat is wrong, or useless.

The thing is, as you have said, stats ARE flawed, and as both you and HGM have pointed out, McCracken's original theory (that pitchers have no ability to control factors that influence the batting average against them on balls in play) is wrong--they do have some ability to influence BABIP, even if it's less than many people think. But that doesn't make the stat totally useless IMO. It's sort of like with Newton's Theory of Gravity and Einstein's Theory of Relativity--we now know that Newton didn't have it exactly right about how gravity works, because in close proximity to a really massive object, the energy of the graivity field itself acts as an additional mass. Einstein showed how it actually works--relativity introduces an adjustment to Newton's equation--but for normal, everyday matters, Newton's theory works just fine.

That's just the simplified Intro to Physics version--I don't want to get into physics debate here--but my point is that a theory can be flawed, but still be close enough to being right that it can produce usable data.

[ohms_law]

The thing is, as you have said, stats ARE flawed,

The statistics aren't flawed though. How can a statistic be flawed, unless you count the events incorrectly or something?

[belial]

The statistics aren't flawed though. How can a statistic be flawed, unless you count the events incorrectly or something?

Ok misleading, based on very poor premises, etc. They are wrong if you try to use them for what the creators intended them to be used.

[belial]

The problem is ohms, pitcher don't totally control walks k and hr's and they do have some control to everything else. A stat that attempts to tell the whole story using just the three things a pitcher can't control anyway is bound to have some results that just aren't worth looking at. Sorry, but it's true.

[HoustonGM]

Ok misleading, based on very poor premises, etc. They are wrong if you try to use them for what the creators intended them to be used.

No, they are wrong if you try to use them incorrectly. FIP is intended to estimate a pitcher's ERA based on the three factors that he has near total control over, and the three factors that are the most stable from year-to-year - strikeouts, walks, homers. Using it for what is intended for is not wrong. You just seem to be making up some other intention to it, which it was never intended for.

The problem is ohms, pitcher don't totally control walks k and hr's and they do have some control to everything else. A stat that attempts to tell the whole story using just the three things a pitcher can't control anyway is bound to have some results that just aren't worth looking at. Sorry, but it's true.

And this is where you're wrong, because FIP is not meant to "tell the whole story." Are you even reading what SirKodiak has linked to? You know, the parts where they say...FIP can't tell the whole story? No stat can. Stop acting like people are proclaiming FIP and other defense-independent stats as the be-all and end-all of pitching stats, because nobody except yourself is. You're fighting a strawman.

[OldFatGuy]

The statistics aren't flawed though. How can a statistic be flawed, unless you count the events incorrectly or something?

How can a statistic be flawed??? Are you serious???

OK, let's say I want to use statistical analysis to gather American voters preference in the upcoming presidential race. So, I go to EVERY single golf and tennis country club in America, and ask their preference for president. I add them up, and not surprisingly find McCain leading Obama 68%-30%.

Now, I know what you're going to say. The stat isn't flawed in that it correctly shows the preference for president at golf and country clubs. But the stat IS flawed when suggesting it represents American voters. We may be getting crossed up in semantics here, where 1 + 1 always equals 2. So, maybe it is more correct to say how you USE a stat is flawed.

But, the above example is a GIGO effect. For the purpose of monitoring American voters, this was garbage in, garbage out. And the same can be said for stats that start out with flawed premises.

[HoustonGM]

How can a statistic be flawed??? Are you serious???

OK, let's say I want to use statistical analysis to gather American voters preference in the upcoming presidential race. So, I go to EVERY single golf and tennis country club in America, and ask their preference for president. I add them up, and not surprisingly find McCain leading Obama 68%-30%.

Now, I know what you're going to say. The stat isn't flawed in that it correctly shows the preference for president at golf and country clubs. But the stat IS flawed when suggesting it represents American voters. We may be getting crossed up in semantics here, where 1 + 1 always equals 2. So, maybe it is more correct to say how you USE a stat is flawed.

But, the above example is a GIGO effect. For the purpose of monitoring American voters, this was garbage in, garbage out. And the same can be said for stats that start out with flawed premises.

Well, you just said it yourself. The statistic isn't flawed, how you're using it is. Statistics themselves can't be flawed unless they are calculated incorrectly. The way statistics are used can be flawed, though. Your poll calculating the preference for president among country clubs in America is perfectly fine, and the result is not flawed. When you prescribe meaning to it that it is not meant to show - what Americans as a whole want - your usage of the statistic is flawed.

[belial]

And this is where you're wrong, because FIP is not meant to "tell the whole story." Are you even reading what SirKodiak has linked to? You know, the parts where they say...FIP can't tell the whole story? No stat can. Stop acting like people are proclaiming FIP and other defense-independent stats as the be-all and end-all of pitching stats, because nobody except yourself is. You're fighting a strawman.

I believe I wasn't talking to you.

[belial]

No, they are wrong if you try to use them incorrectly. FIP is intended to estimate a pitcher's ERA based on the three factors that he has near total control over, and the three factors that are the most stable from year-to-year - strikeouts, walks, homers. Using it for what is intended for is not wrong. You just seem to be making up some other intention to it, which it was never intended for.

Except he doesn't have near control over them and he has more control than he is given credit for over the other outcomes. It's basically worthless to draw any conclusion over. To use only those 3 at all is using them incorrectly.

[HoustonGM]

This is a public conversation. If you want to talk directly to somebody with no input from anybody else, use private messages.

[belial]

Yeah I guess when i addressed ohms I really meant for someone else to answer. How ignorant of me.

[HoustonGM]

Except he doesn't have near control over them and he has more control than he is given credit for over the other outcomes.

What do pitchers have vastly more control over - walks/strikeouts/homers or hits? The correct answer is walks/strikeouts/homers, and every statistical study done in the history of the sport will back that up. The simple fact of the matter is that walk rate, K rate, and HR rate are very stable, while hit rate fluctuates a lot. Walk rate, K rate, and HR rate are exclusively between the batter and the pitcher. Hit rate is between the batter, the pitcher, and the fielders. It's simple common sense that a pitcher has a huge amount more control over those peripheral stats as compared to hits.

It's basically worthless to draw any conclusion over. To use only those 3 at all is using them incorrectly.

And once again, I suppose you're overlooking the multiple times in this thread and the articles linked from this thread where everybody is saying that you CANNOT EXCLUSIVELY USE THEM. You're arguing against something nobody is saying.

[OldFatGuy]

Well, you just said it yourself. The statistic isn't flawed, how you're using it is. Statistics themselves can't be flawed unless they are calculated incorrectly. The way statistics are used can be flawed, though. Your poll calculating the preference for president among country clubs in America is perfectly fine, and the result is not flawed. When you prescribe meaning to it that it is not meant to show - what Americans as a whole want - your usage of the statistic is flawed.

All right, I guess it is a matter of semantics, and my defensiveness. My apologies to Ohms, I guess my initial sentence in that post was a bit defensive, but I won't edit it out even though I'd like to.

But, speaking of semantics, would you agree or disagree with the following.

"Based on the example above, that stat is flawed for determining American voters attitudes in the upcoming Presidential election."

[HoustonGM]

But, speaking of semantics, would you agree or disagree with the following.

"Based on the example above, that stat is flawed for determining American voters attitudes in the upcoming Presidential election."

I'd agree with that, for the most part. But, it's the usage of the stat that makes it flawed. Using batting average to determine a player's power is a flawed usage of batting average. It doesn't make the stat itself flawed though.